Hi everyone,
I’ve got a model (see attachment) where there’s an “area of interest” (ship) which is partially obscured by unwanted elements (red stamps, iron poles, etc). I’m looking for the best method to create an orthophoto of just the ship (without it containing all the unwanted elements).
- First method that comes to mind is to simply delete all unwanted geometry in external software, reimport the mesh, texture it, build orthos. This would be the easiest method, but I fear that by removing the unwanted geometry, the texture of that unwanted geometry will instead be projected onto the hull of the ship, which we of course don’t want.
- Second option would be to do the same as the first option, and then mask out all unwanted elements in each picture, but that would be insanely time consuming so really it’s not a viable option.
- Third option would be to use the current geometry (including unwanted geometry), texture it at insanely large texture size, export the mesh + huge texture, edit the model in external software in order to remove unwanted components, and then simply make renders of the model w remaining texture to serve as orthophotos. However I fear that my computer won’t be able to handle the necessary texture size + that the quality of the results would still be less than actually building an ortho using the standard workflow (intended resolution: 40 000 pixels wide).
- Finally I could set my “ortho bounding box” in RealityCapture, and create dozens of individual orthophotos of the same area, but progressively moving the bounding box closer and closer to the ship. I could then import these orthos onto separate layers in fi Photoshop, delete unwanted parts from each layer, and merge the resulting layers which would only contain the ship. However the number of orthophotos I’d have to make would be ridiculous, since some of the unwanted elements are slanted, and since the shape of the hull is also slanted.
Any advice on what the best workflow would be? Perhaps some workflow I haven’t even considered yet?
All input is greatly appreciated as always 
Thank you,
Thomas